Testing, Evaluation, and Remediation
What to know
Accessibility testing combines automated and manual methods. Automated tools (axe, Lighthouse, WAVE) catch a subset of issues - missing alt, contrast, labels - but cannot judge meaningfulness or keyboard flow. Manual testing includes keyboard-only navigation, screen reader testing (e.g., NVDA + Firefox, VoiceOver + Safari), zoom to 200%, reflow at 320 CSS px, and reduced-motion checks. User testing with people with disabilities provides the strongest evidence.
Remediation should prioritize by severity and user impact: blocking issues (keyboard traps, missing labels, inaccessible forms) first, then contrast and semantic issues, then polish. Document findings with WCAG criterion, severity, and reproduction steps. Regression testing prevents reintroducing fixed issues.
Traps
- Treating a clean automated scan as proof of conformance; it is not.
- Testing only with one browser/screen reader combination; behavior varies.
- Confusing "no errors" with "usable" - a page can pass automated checks and still be unusable.
- Forgetting to test dynamic content, modals, and error states.
Rule of thumb
Use the "three-legged stool": automated scans + manual keyboard/screen reader checks + user testing. Report each finding with the WCAG criterion and a reproducible step sequence.
Worked example
A checkout form passes axe but a keyboard user cannot reach the "Place order" button because focus is trapped in a modal. Automated tools miss this; manual keyboard testing reveals it. Fix: correct focus management, then add a regression test.